This one is an opinion piece - you have been warned.
There are a lot of companies, organisations and people at NAB crowing about various kinds of colour processing for cameras and postproduction. We had the Academy and ACES, Arri are shouting about some sort of new setup for the Alexa, and even on the Sony booth it's possible to find a Resolve grading station and flick through various 709 and ACES-based transfer functions on some F65 material.
Where this is about calibrating displays for consistent results, as it is with things like Filmlight's Truelight and most of the ACES system, this makes perfect sense: there's not much point in encoding images using a tristimulus RGB representation if we don't know which red, green and blue we mean. To that extent, some of the most promising developments in cameras and displays involve increasing color gamut, particularly by deepening the notoriously pale and yellowish Rec 709 green. Promisingly, OLED displays (about which I've raved enough already) are manufactured by a process which makes primary selection fairly flexible, so with any luck we'll be able to have imaging systems soon - other than film - which are capable of making a dark green tree look anything other than sickly.
But where improved "color science" - term as populised by Red - is being promoted as a major discovery, I'm not so sure. We refer to cameras having "log output" as if that really means anything. Even if the data coming off the sensor were linear - which it is very far from being - most of the manufacturers seem to feel free use more or less any transfer function they choose, any curve that makes the camera look good. Which is fine, but it's largely a matter of opinion. Don't, as some manufacturers are doing here in Vegas, tout your latest sensor-to-recorder transfer curve as some sort of major discovery to the benefit of mankind. If a new curve gives you greater dynamic range, well, shouldn't you just have done it like that in the first place? What's the situation, are these building cameras for an industry which greatly values dynamic range, then deliberately cropping-off highlight detail because that's what a hopelessly outdated spec says to do? That's a fairly appalling engineering decision, if so, especially if two years later you rectify that mistake and try to claim it's a major bit of R&D.
NAB: Stop crowing about curves
No replies to this topic